Introduction

Over the last three decades, public sector reforms have at the face of it aimed at giving public organizations more autonomy from governments. Placing public organizations more at arm’s length of direct control from political authorities has been advocated as a way of improving performance in service provisions. It has been widely discussed how democratic governments are changing their relationships with public service organizations, what the implications for their external and internal control might be, and what such changes deliver in terms of expected improvements of public services (Pollitt and Bouckaert 2000; Christensen and Lægreid 2006; Olsen 2009). ‘Autonomization’ has thus been part and parcel of the wider debates about shifts from government control to a regulatory state (Majone 1996) and from ‘Old Public Administration’ to ‘New Public Management’ (Dunleavy and Hood 1994). Higher education in Europe has been no stranger to such reforms and the question of the autonomy and control of the university is of significant scholarly and policy interest.

The main aim of this paper is to contribute to the conceptual and empirical understanding of governmental control and organizational autonomy in higher education studies. Our approach is to link up higher education studies to the wider scholarly literature on public management reform, principal-agent relationships, and sociological approaches in the study of public policy and institutional design. First, we ask: what is the meaning of autonomy in the narrative of change of contemporary higher education policy? In order to understand the historically contingent and changing nature of the concept, we begin our analyses by looking at the shifting meaning of autonomy from the past based on institutional trust and professional autonomy to the present based on organizational autonomy. We point to contextual and political factors accounting for the reformulation of autonomy within the field of higher education. Second, how can we empirically assess the effect of political reforms for autonomy and control of universities as organizations? We distinguish between formal and real organizational university autonomy and develop a multi-dimensional taxonomy for its empirical study. Our taxonomy is applied to the Dutch case that highlights our theses of ‘regulatory autonomy’, that is the uses of organizational autonomy as a tool of governmental regulation within a new regime of governmental control. The last research question posed in this article is: What empirical evidence do we hold for a link between university autonomy and performance? In this section, we turn to international studies on organizational autonomy and performance in higher education, and discuss their findings and limitations. In the conclusion, we summarise our findings and point to promising future avenues for studies on organizational autonomy and performance as two core concepts in contemporary higher education debates.

The changing idea of university autonomy

At the very heart of the contemporary debate about university reform and state-university relations is the question of university autonomy. As “autonomy is contextually and politically defined” (Neave 1988: 31) the meaning of autonomy changes over time. The purpose of this section is to analyse a major shift in the idea of autonomy that is underpinning much of the political reform of the university. We argue that the dominant narrative of reform implies a shift towards the ‘organizational autonomy’ of universities as emerging actors in the field of higher education, and point to contextual and political factors which account for the reformulation of university autonomy.

When considering if and to what extent universities might have become more autonomous actors, it is obviously important to clarify what is meant by this term in general as well as in the context of university reform. In its classical meaning autonomy refers to ‘auto’ (self) and ‘nomos’ (rule or law) and ‘autonomia’ refers to the right to self-government, e.g. to a state free to determine the rules and norms by which it wants to live (Ostwald 1982, in Ballou 1998: 103). Key for our understanding is thus that autonomy refers to both the actor’s self (having ability or capacity) and the actor’s relationship to its environment (independence or freedom from external control). Historically, the right of the university to govern itself has always been an important though contested issue (Perkin 1991) and was intimately tight to the issue of the professional autonomy of academics (Enders 2006), i.e. their role and powers in the self-governance of the university as well as their academic freedom to pursue teaching and research without fear of intervention or punishment. The modern conception of the university embraced the idea of the university as a distinctive social institution which deserves special status in terms of autonomy and academic freedom based on a ‘social compact’ that evolved between higher education, the state and society. The belief that the university requires autonomy from substantial political or corporate influence to function optimally was in turn linked to the role of the state as the guardian of the university in substantive matters, guaranteed state funding, at least in continental Europe, strong professional self-governance and protection of academic freedom.

This idea of the autonomy of the university was reflected in the idealist tradition of the modern university, most namely in the Humboldtian model, as well as in influential conceptions of the university in the social theory of modernity. In the functionalist tradition, Parsons theorized the emergence of the autonomous modern university as part and parcel of the social differentiation of modern society, the emergence of a functionally superior setting of societal sub-systems and their related professions (Parsons and Platt 1973). In a related though different influential conception build into the Mertonian sociology of science, the autonomy of science called for a limited or protective role of the state viz-a-viz the university as the dominant organizational host of scientific discovery. Bourdieu’s (1988) field concept theorized higher education as a field with a high degree of autonomy in that it generates its own organizational culture consisting of values and behavioral imperatives that are relatively independent from forces emerging from the economic and political fields. For Bourdieu, the relative autonomy of science as well as the role of higher education in the social reproduction of society and the related proximity of the university to societal elite legitimate its autonomy. Bourdieu’s framework thus “emphasises the autonomy of the field not as a given product of consensus but as a dynamic product of a permanent conflict within a mediating context linking the university to the external environment” (Naidoo 2004: 467). Accordingly, autonomy varies from one period to another, from one field to another, and from one national tradition to another.

In the continental European model of higher education (Clark 1983), the development of the modern university and of the academic profession was intimately tight to the emergence of the modern nation-state, its political institutions and bureaucratic apparatus. State-university relationships developed a specific form characterized by a strong role of the state as well as of the academic profession. On the one hand, the state functioned as a strong regulator and funder of universities exercising bureaucratic control over procedural matters. On the other hand, the state protected the autonomy of the university as a social institution, academic freedom as well as academic self-governance, and substantial matters were delegated to academics within a broad state framework. The protection of academic freedom and guaranteed funding by the state enabled universities to establish normative and constitutive principals without being subject to strong external design (Olsen 2007). The university as an organization developed, however, very limited authority and capacity to regulate itself, e.g. in terms of its organizational boundaries, related entry and exit requirements, its size and shape, its broader character and functions. In consequence, universities were ‘bottom-heavy’ with low potency for collective action. Organizational leadership was weak compared to other organizations. Organizational change took place mainly through continuous local adjustments, while major change was difficult to achieve. Central organizational policies were often weak and interventions on this basis may have only minor, local effects (Weick 1976). It is the academic professionals who act in the university as an arena, rather than the university itself as an organizational actor (de Boer et al. 2007).

The new autonomy of universities

Over recent decades, this ‘social compact’ has been eroding, and the special status of the university as a social institution is no longer taken for granted. In many OECD countries and beyond, higher education reformers developed measures within the context of wider reforms of public services and public management (Hood 1991; Pollitt and Bouckaert 2000; Olsen 2009) in order to change the coordination of universities. Shifts in the way of regulating and funding universities reflected the drive towards mass higher education as well as the overriding concern on the appropriateness of the output of higher education and research for economy and society (Naidoo 2008). Higher education had become bigger, more expensive, less elitist, politically more visible and economically more strategic. “This is what occurred in many countries over the last decades. As the higher education system massifies, so these external and governmental pressures on the higher education subsystem may be expected to increase. Within the knowledge based economy, the connection between the higher education institution subsystem and policy goals of economic growth become sharper.” (Ferlie et al. 2009: 9). This is for instance clearly echoed in the communiques of the European Union in the last decade. Models of state supervision, instead of state control, of output control, instead of process control, of ‘market-like’ competition combined with attempts to strengthen the actorhood of universities as organizations have become prevalent.

This new idea of how to organize government-university relationships and the autonomy and control of universities has been inspired by the growing popularity of New Public Management approaches that find some of their theoretical background in principal-agent theory. Three main assumptions are characteristic for modeling government-university relationships in terms of principal-agent theory. First, and sharing some basic foundations with economic neo-institutionalism (Furubotn and Pejovich 1974; Jensen and Meckling 1976; Pratt and Zeckhauser 1991; Moe 1984), the framework assumes that rational actors strive to maximize their preferences according to their priorities. Given that the priorities of the principal (the government) and the agent (the university) may differ problems can arise due to a separation of interests. The second assumption is that there is a problem in the asymmetry of information between those who command (the government) and those who receive commands (the university). Agents may thus use their discretionary room stemming from this asymmetry to pursue their own interests and will not necessarily act in accordance with the goals of the principal. As a result of that, agency slack, defined as unwanted independent action by an agent, is a potential problem which requires action from the principal. The third assumption of principal-agent theory is that the rational design of institutions will help to overcome these typical problems in the relationship. On the one hand, devolving authority from government to the university (autonomy) is beneficial for government for various reasons. Universities hold expertise and information not easily available to the government; universities can exercise internal control with closer proximity to their primary activities and products; and governments can turn over some risks to the universities. On the other hand, typical relationship problems arise because of the self-interested ‘opportunism’ of the university as an agent. Devolving authority from the government to the university should thus go together with incentive structures to control the university. Examples of such control mechanisms are specific rules (that limit discretion), monitoring university behavior and performance, bonding universities in performance contracts, creating competition between universities.

Principal-agent theory thus models the universities’ autonomy as part and parcel of a new regime of control. The strengthening of universities as strategic organizational actors with capacities for managerial self-regulation and internal control, and new tools of governmental control are expected to increase organizational performance.

Across continental Europe, we can see the influence of such ideas about changing principal-agent relationships between government and universities. Many governments have announced attempts to step back from the traditional regime of state control and to take a more distant framework setting role in the relationship with their universities. More efficiency and effectiveness in university provision is expected by delegating some authorities to universities as more autonomous and, separately, accountable organizations. Such changes also have supported the view that universities should become more managed enterprises (Bleiklie 1994). Through stronger executive leadership and professionalized management structures strategic capacities of universities as organizations should be strengthened. Another key element of the reforms is to increase the level of competition between universities, to allocate resources according to the performance of the organization as a whole, and to shift the focus of state control from procedural matters to output-related monitoring of performance (accountability).

A new narrative about university autonomy has thus been emerging in which the university as an organization has become an important focus of attention in the system’s coordination. Universities are supposed to act as social entities that possess a certain degree of autonomy and sovereignty, with self-interested goals as well as with rational means, commanding independent resources and visible boundaries (de Boer et al. 2007).

The grand narrative noted above which finds some inspiration in principal-agent theory appears to give a standard recipe for governing universities better and improving their performance. However, we need to treat this with caution given competing theoretical perspectives deriving from a sociological institutionalist approach in the study of public policy and institutional design (Christensen 2009; Christensen and Lægreid 2006; Egeberg 2003). According to this approach, grand narratives of changing government-university relationships might very well provide an important input into actual reforms. They represent changing belief systems about what appropriate and efficient ways of organizing this sector might be. If they become internationally taken for granted, pressure will grow to engage with related reforms and practices. What will be done, however, is not necessarily what has been said, but an interpretation and adaptation of dominant models.

A main assumption of this approach is thus based on the concept of bounded rationality (March and Simon 1958). Bounded rationality implies, first, that actors have limited attention and information; they cannot address all alternatives and consequences. Reform and response will thus not be the outcome of a near perfect rational choice but will be limited in its capacity to deal with complex realities. Second, historical and cultural traditions in organising government-university relationships will come into play as well. Path dependencies might constrain what is possible and the reform road taken reflects some historical legacies and sector-specific trajectories. “Build in here is also a negotiation perspective, suggested by Cyert and March (1963) and March and Olsen (1983), stressing that public decision-making processes might result from negotiations and compromises between actors with different interests.” (Christensen and Lægreid 2004: 11). Third, norms and values that have been institutionalized within the organizational field of universities interact with reforms aimed at changing the autonomy and control of the university. Compatibility of reforms might be important for understanding their success and failure. Change will be difficult if it does create disruptions undermining the legitimacy and trust in the university as an institutionalized form. Fourth, changes in the formal autonomy situation of universities do matter but there are good reasons to assume that autonomy in practice is not a perfect copy of formal autonomy. Formal rules for autonomy might be implemented or not, and they cannot prescribe in advance practices in universities. Finally, characteristics of the focal organizations—such as the complexity of their tasks (and thus the measurability of their outputs), or their size and age—might influence reform trajectories as well as their effectiveness. Also in this respect, universities as organizations are certainly an interesting case characterized by unpredictability of results and task ambiguity, and by performance being based on professional expertise and recognition in inter-organizational professional communities.

Institutional design for public sector organizations—such as the issue of the autonomy and control of universities—will be the outcome of a blend of factors such as changing political beliefs and goals, path-dependencies and political compromises, and organizational characteristics. These theories are skeptical when it comes to a standard model for success in higher education.

In the following sections, we will thus go into ‘practice’ in understanding organizational autonomy and control in higher education and the related issue of university performance. First, we will focus on the Dutch case as a particular example of higher education reform and the empirical mapping of Dutch universities’ autonomy situation. Subsequently, we will discuss international studies on the relationship between university autonomy and performance.

Formal and real university autonomy: the Dutch case

Obviously, the grand narrative of change discussed above addresses the autonomy of the university as an organization in a new way. In the following, we want to discuss the autonomy situation of Dutch universities—a case of particular interest given that Dutch higher education policy has been a continental European front runner of reform in this sector. From a bird’s eye view, key words of the governmental reform agenda seem to suggest changing principal-agent relationships between government and universities as well as a ‘new freedom’ of universities: Deregulation and devolving authority from the government; promoting self-organization of the sector; shifts from detailed ex ante control to ex post performance monitoring; performance-based funding and more competition; strengthening the position of the university as an organization and granting more autonomy have all been on the governmental reform agenda. The generic character of the reform narrative makes it at the same time difficult to empirically assess the new autonomy of the university.

In doing so, our methodological approach builds on the work of Verhoest et al. (2004) on measuring organizational autonomy of public sector organizations that we have applied to the world of higher education (de Boer et al. 2010). Following Verhoest et al., we consider two kinds of organizational autonomy: autonomy as the level of the universities’ decision making competencies and as the exemption of constraints on the actual use of such competencies. Universities’ decision-making competencies has four dimensions: Managerial autonomy addresses discretion over financial matters as well as human resources management and management of other production factors (logistics, housing, and organization). Policy autonomy indicates the extent to which a university can take decisions about the quantity and quality of the goods and services to be delivered, and the target groups it wants to reach. Governance autonomy concerns the extent to which universities can decide on internal structures and processes, and policy instruments.

The second kind of autonomy refers to the absence of constraints on the organization’s actual use of its decision making powers. Financial autonomy deals with the universities’ dependency on governmental funding and alternative sources of income. Legal autonomy relates to the legal status of the organization (e.g. public or private) and its implications for the autonomy situation. Interventional autonomy refers to the extent to which the university is free from reporting requirements, evaluations or audits and from possible threats of government intervention.

This comprehensive taxonomy fits the two main characteristics of autonomy discussed earlier (the capacity to act and the freedom from external control) as well as the multi-dimensional and situational concept of autonomy. The taxonomy is in line with our focus on the organization and integrates issues of autonomy and control (e.g. via accountability) that usually tend to be separated in higher education studies. The focus on the organizations points, however, to a major limitation of the taxonomy that has little to say about professional autonomy, especially aspects of academic freedom are not covered. The taxonomy covers basic distinctions (such as procedural and substantial autonomy) as well as integrating important sub-dimensions (such as organization/internal governance, finance, personnel and academic matters) of earlier studies on university autonomy (Anderson and Johnson 1998; Berdahl 1990; Estermann and Nokkala 2009; Moses 2007; Levy 1980; Ordorika 2003). The outcomes of our application of the taxonomy to the formal autonomy of Dutch universities are presented in the next section.

Formal autonomy of Dutch universities

With respect to human resource management Dutch universities have certainly gained in autonomy because the government has been stepping back from detailed rules and regulations. Dutch universities are nowadays the employers of all their staff and can introduce line management instruments. They can decide themselves on the number of academic posts and select persons of their choice without governmental interference. The university autonomy remains, however, restricted due to other actors and mechanisms. Salaries and many other labor conditions are set by collective bargaining between the national employer association and the unions, followed by local negotiations between the individual university and the local unions. The Collective Labor Agreements at the national level, legally binding for the universities, comprises among other things agreements about minimum and maximum salaries, detailed descriptions of staff positions and career ladders, job appraisals and procedures for promotions. Managerial discretion is furthermore bound to the general legal rules for hiring and firing in the Netherlands and the Dutch public service. Managerial discretion has increased but is still being limited within the framework of national rules and regulations.

The university’s decision-making capabilities to manage financial affairs are considerable. Dutch universities can decide on how to spend the public operational grant, set the tariffs for contract activities, borrow money form the capital market, build op reserves and carry over unspent resources from 1 year to the next. They can set up and change rules for internal budget allocation, develop their own managerial tools for internal performance agreements or targets for budget allocation or organize internal competition in contests for certain parts of the budget. Dutch universities can, however, not decide on tuition fees for Bachelors and Masters programs: they must charge tuition fees that are fixed by the ministry (€ 1.732 annually).

The level of policy autonomy is relatively moderate to low. Dutch universities have to accept all qualified Bachelors students (with exceptions for some disciplines), although they may ask the minister to fix the number of study places for particular programs because of huge demand in combination with limited capacity (so-called ‘capacity fixus’). Universities can select their Masters students and as the result of that determine themselves the number of study places for Masters programs (again, with some exceptions). The development of new Bachelors and Masters programs is subject to both accreditation (by a national agency) and ministerial approval. Research programming is up to the university. However, research programming by governmental actors, increasingly including the ministry of economic affairs as well as the research council, tries to set incentives for organizational choices on preferred topics.

The level of governance autonomy is relatively moderate to low as well. The universities’ governance structure is to some extent prescribed by law. By means of university bylaws the universities have some leeway to develop a structure to their taste, but within the legal framework. Although appointed by the university’s supervisory board, the national Act stipulates the guidelines for the selection of the rector. The members of the supervisory board, all external, are appointed by the minister. National law also prescribes a further top-down decision-making on managerial positions: the university leadership appoints the deans and directors of central university research institutes, the deans appoint head of departments etc. Traditional bodies of academic self-governance have been abolished or have been disempowered by having a consultative function mainly.

As regards legal autonomy, the situation of Dutch universities has not changed. They are public agencies bound to the general rules and regulations that apply.Footnote 1 The financial autonomy situation of Dutch universities has somewhat changed due to a modest trend towards revenue diversification. Roughly two-third of the budget of Dutch universities derives nowadays from the government. Universities are seriously dependent on this income stream that comes with strings attached as regards funding by student numbers and a related component to subsidize basic funding for research. Some shift of governmental money for research to the research council and other bodies running governmental programs for research has, however, led to some revenue diversification. This implies that Dutch universities are nowadays somewhat less dependent on direct governmental funding while other actors (intermediary bodies, other ministries) have gained control over some part of the funding of the university.

Finally, the interventional autonomy of Dutch universities is low since they face several ex post reporting requirements. They must have internal and external evaluation systems for both teaching and research. They have, across the board, the opportunity to decide on the methods they want to use, but these methods are evaluated by a national agency. They have to report on their activities in annual reports and audited financial statements. Teaching programs are due to accreditation by an intermediary agency. In addition, external research sponsors have established their own reporting requirements that might range from rather loose procedures for research council grants to detailed prescriptions by sponsors such as the government itself or the European Commission. This adds another layer of control on the university.

This brief sketch already highlights the dynamic and ambiguous nature of the formal autonomy situation of the Dutch university that would be blurred in generic statements about the ‘new freedom’ of universities. The balance of autonomy and control in relation to the government has certainly changed. Overall, managerial autonomy in finance and human resources has increased while new strings have been attached in parallel to strengthening leadership and management in universities. Political and governance autonomy are still rather limited. Legal autonomy has not changed and financial autonomy from government is still low as well. Interventional autonomy has certainly decreased.

Further, the rhetoric surrounding Dutch university reform stresses their increased autonomy while the government systematically down-played old and new forms of control. The government is still very present in defining the rules of the game. Shifts towards more formal autonomy of universities have been initiated by the government, in some cases against the opposition of the universities, and government is controlling the autonomy of the university. What has been given can also be taken away if the government would decide to row back in its overall arrangement of governing universities.

Finally, the Dutch government has placed universities in a more complex multi-level and multi-actor constellation of autonomy and control. Some control has been delegated to collective bargaining; intermediary bodies have taken over some of the traditional governmental roles in rule setting and funding; new governmental actors have entered the scene, and government appoints external board members of universities being expected to control major aspects of financial policies, human resource management policies, and strategic development of universities.

Formal and real autonomy: a contested relationship

The previous section has led us to a number of conclusions as regards the multi-dimensional empirical investigation of the autonomy of universities and the current formal autonomy situation of Dutch universities. Arguably, an understanding of the formal autonomy of universities goes beyond traditional perspectives focusing on the legal status of public organizations only. Formal autonomy is, however, only one side of the coin. Further understanding of the real autonomy of universities is needed as well.

From the point of view of a structural-instrumental perspective these two sides of the coin would be expected to look pretty similar since the real autonomy of an organization should reflect its formal autonomy. As we have argued earlier, there are conceptual arguments for expecting that the degree of formal autonomy granted to universities does not necessarily translate into the same degree of real autonomy.

Government might to some extent have stepped back from control of its universities while it still remains the most important player in the field in terms of rule-setting and funding. Nowadays, Dutch universities find themselves in a situation of ‘regulatory autonomy’ that has severe implications for the inter-linkages between autonomy and control. Universities work in the shadow of hierarchy, and the Dutch government has various times shown its interventionist capacities when universities tried to test the boundaries of their ‘new freedom’. Under such circumstances, universities will try to anticipate the political and administrative position and up-coming policies, which is a well-known mechanism from the study of decision-making within the ministerial bureaucracy (Mayntz and Scharpf 1975). As a consequence, formal autonomy of the university will be higher than the actual level of autonomy used because university leadership anticipates the government’s position. More importantly, Dutch autonomy policies interact with other new regulatory policy instruments such as output funding, performance monitoring and (bilateral) contracting. Universities have been granted more autonomy in financial and human resource matters while new instruments attempt to steer organizational choices and behavior more effectively. It allows the government to force universities under the pretext of voluntary strategic choice into certain directions desired by the government, which would not have been possible with old policy instruments. Under these circumstances, and in line with propositions derived from principal-agent theory, organizational autonomy becomes part and parcel of a new regime of control of universities. Along the same lines, Dutch government is currently preparing further bonding of universities via performance contracts.

University reformers might experience an implementation gap, and the university is not an unlikely case for experiencing problems in actually implementing a new regime of autonomy and control. One reason for this is that universities have traditionally been different from many other organizations (loose coupling between administration and academic core, high professional autonomy, weak leadership and management). Since the mid-1980s, the Dutch were a continental European frontrunner in replacing the previous centralized arrangements with an autonomy policy designed to grant universities increased decisional leeway (Goedegebuure and Westerheijden 1991). Such radical change was, however, met by skepticism from the universities. The result was that the policy impact on organizational autonomy has not been as great as expected. The traditional approach to policy-making, based on consensual and distributive policy, continued to prevail (de Boer and Huisman 1999). It was in 1997, that the internal governance system of universities was changed by law in order to force Dutch universities to develop managerial autonomy and intra-organizational hierarchical steering capacities according to the logic of the verticalization of responsibility for decision-making.

Universities experience more autonomy in some dimensions while stronger (state) control is exercised in other dimensions. Given the inter-dependencies between dimensions of autonomy this might result into paradoxical effects for the organizational autonomy of the university. In the Dutch case, managerial autonomy for budgeting in universities has, for example, certainly considerably increased. At the same time, state-induced new control technologies for financial auditing have effected in streams of rule setting and red-tape for the organization that are likely to limit managerial discretion in this area. Managerial autonomy for human resources has increased as well while traditional state control in this area has partly been delegated to collective bargaining. Collective bargaining has resulted into an impressive amount of binding rules and regulations that severely restrict the decision-making capacity of the organization in this area. In addition, in areas where the state relaxed control universities themselves have stepped in and created new rules and mechanisms for internal control. Micro management of the government is replaced by micro management of the university apex. Such observations point into the direction of a Tocquevillian paradox that Hood and Peters (2004) noted in their review of public management reforms, namely the production of bureaucratic activity of a style even more rule-based and process-driven than the traditional forms of public bureaucracy that were meant to be supplanted.

The Dutch case also points to the usefulness of a task-specific perspective. We refer here to the measurability of outputs of organizations that forms an important pre-condition for performance monitoring and bonding according to principal-agent models. Wilson (1989) has classified public organizations along this dimension and developed ideal types of public organizations confronted with specific control challenges. Arguably, universities outputs are not as easily accessible as the outputs of organizations that Wilson called ‘production agencies’. The Dutch response has been to invest into ever growing amounts and specifications of outputs and related indicators derived from the universities themselves as well as from system monitoring. Obviously, the increasing efforts to make the activities and outputs of academic work more observable, measurable and comparable reflect more general changes in political dispositions as regards principal-agent relationships between government and university. They also stimulate closer performance monitoring within the universities. Related performance funding incentivizes universities to make strategic choices aligned with government’s goals. Subsequently, government struggles, however, with two unwelcome consequences of success in performance steering. First, standardization of performance monitoring and funding leads to similar choices of universities undermining organizational specialization within the system (or profiling of universities). Accordingly, government tries to implement performance contracts including some premiums for profiling activities of universities. Second, success in incentivizing growing quantitative outputs raises concerns about the quality of products delivered. In consequence, further efforts are made to monitor quality and to sanction deviant behavior that fuel the rise of internal and external audits and evaluations.

Finally, Dutch reformers might have under-estimated the problem of the specific nature of the university as an organization for internal managerial self-regulation and control. Three decades of reform confirm that characteristics of the primary tasks of the university strongly affects the potentials and barriers for managerial autonomy and control. Musselin (2007) and Whitley (2007) have provided arguments to assume that universities’ capacities to develop firm-like strategic actorhood will be limited even under conditions of decreased state control, increased organizational autonomy and growth of internal managerial control and surveillance. Such limitations are ascribed to inherent characteristics of universities as part of the public science system, i.e. the inherent technological uncertainty of their core activities and their deeply embedded fragmentation. As organizations, universities thus possess limited discretion over expected results. Further, systematic planning, coordination and integration are limited by the division of academic labour along the lines of fields of research and teaching with their specialist knowledge and skills. Central coordination and integration for collective goal achievement will have to rely on the contributions of loosely coupled parts of the organization to the whole. Under such conditions “strategic choices are more similar to those of holding companies and investment portfolio managers than entrepreneurial decision-making in more authoritatively integrated and directed work organizations” (Whitley 2007: 25).

Dutch higher education reform re-visited

Our discussion of Dutch reforms highlight the ambiguous nature of reform as well as the influence of principal-agent thinking. After the introduction of a new steering philosophy entailing a new principal-agent relationship, government revised autonomy and control of universities as an attempt to align changing political interests and organizational action. Organizational autonomy was increased in some ways while existing control mechanisms were adapted and new ones were introduced. Specific rules that curtail university choice (e.g. organizational structure and boundaries, legal status and financial dependencies) have been maintained. Managerial capabilities for internal self-steering and control have been strengthened (financial affairs and human resource affairs). Accountability requirements to monitor agent behavior have been increased (quality assurance schemes, evaluation and accreditation, performance monitoring). Incentives have been set to increase competition and to control the flow of resources according to performance. Such instruments allow the government to incentivize universities’ behavior into certain desired directions, which would not have been possible with old policy instruments. Another more recent step concerns attempts at contracting relationships between Dutch government and universities. Contracts provide another instrument to align the agent to the goals of the principal and might provide more security to the agent as regards expectations and funding. We also witness a change in the principal structure itself from a dominant single principal (the government, namely the ministry of education) to multiple principals (the government, increasingly including other ministries, accreditation and funding bodies, evaluation agencies).

At a first glance, Dutch higher education reform seem to have followed a master plan that provides a clear break with the regime of the past. Our analyses points, however, at reform dynamics that support expectations derived from sociological institutionalist approaches in the study of public policy and institutional design. Over three decades, the Dutch government has been struggling with implementation gaps, the persistence of traditional modes of consensual and distributive policies, resistance within universities resulting in legal enforcement of change, the perennial problem of the measurability of universities’ outputs, and limitations to a clear cut hierarchical regime of internal control in universities. The current political compromise reflects changing political beliefs and goals as well as politico-administrative path-dependencies and compatibility problems within the field. We find support for the theses that neo-corporatist traditions as well as path dependencies of the Rechtsstaat have constricted the possibilities for more radical change (see also, Westerheijden et al. 2009). The formal organization as well as the culture of universities have been quite alien to a reform agenda that has been partly inspired by the idea that universities might be treated as production organizations. And imperfect outcomes and unintended effects of policies of the past fuel on-going streams of new policies.

After three decades of reform and re-reform a new regime of governmental control and organizational autonomy has, however, emerged. Dutch universities have gained in managerial autonomy while they have lost in institutional autonomy. Autonomy policies for strengthening the internal control of universities and their actorhood as organizations are combined with regulatory policies for external control that constrain organizational choices and are expected to align universities more closely with governmental goals. When government retreated from traditional forms of control in favor of self-regulation, it turned towards control through different regulatory instruments and actors including the university as an organizational actor itself. Re-regulation has contributed to the emergence of a multi-level and multi-actor playing field of organizational autonomy and direct or indirect governmental control. Government has also repeatedly emphasized its interventionist capacities in cases of systemic failure or low performance. Universities work nowadays in the shadow of a complex hierarchy that aims at getting the incentives right in order to control and improve their performance.

University autonomy and performance: looking through a glass darkly

As we have discussed earlier, university reform including the ‘autonomization’ of universities is claimed to make deliberate changes to the structures and processes of universities with the objective of getting them (in some sense) to run better. There is thus growing interest in and use of organizational and system performance information in higher education. Performance is becoming the ultimate ‘dependent variable’, particularly because universities are increasingly considered from an instrumental perspective. According to this view, support for the higher education system and its organizations can no longer be based on institutional trust but depends on their contribution to society and economy (Olsen 2007).

Consequently, and as we have exemplified by the Dutch case, higher education policy gets instrumented by measures for output control, new accountability requirements, quality assurance systems, performance-based funding mechanisms or the use of contracts that call for related performance information. The new rules of the game are expected to set incentives for universities to improve performance that trigger a need for performance-related information in order to identify strengths and weaknesses as well as strategic options.

While the attention for and use of performance measures in higher education has grown, little political action has so far been taken to evaluate the impact of changes in systems’ governance and organizational autonomy on the performance of universities and higher education systems. Research in organizational studies has focused on regulatory public agencies and shown a mixed picture of positive and negative correlations between organizational autonomy and performance (see Verhoest et al. 2010). Empirical investigations into university autonomy and performance are a quite recent phenomenon. In the following section, we will present and discuss their contributions to the debate and further hypotheses building as well as their limitations.

Organizational autonomy and performance: previous studies re-visited

Research on university autonomy and performance has first emerged in the United States in an era of declining public support for universities combined with increased state-wide regulation and control. Various studies investigated the relationship between governance, autonomy and performance paying special attention to the similarities and differences between public and private universities, and the different systems of governmental oversight of their state universities by their board structures. Knott and Payne (2004) have reviewed previous studies that pointed to superior performance of more autonomous private universities compared to public universities especially in research. As regards public universities, Volkwein and Malik (1997) found no relationship between academic and financial autonomy on the one hand and universities’ performance on the other hand. Instead, size and wealth of the state and of the university correlated with research performance. Eykamp (1995) found financial autonomy as the most important autonomy predictor of performance. In addition, constitutional autonomy, speed of appointments made to the board by the governor, and campus size were found to be positively correlated with performance. Knott and Payne (2004) looked as well into the impact of state governance structures, the autonomy and control of public universities, and their performance. The systems’ governance structure influences important choices in terms of resource allocation and sources of revenue. The governance structure may also have an indirect effect on how organizational management makes decisions: in highly regulated structures more attention is paid to politically prioritized issues such as lower tuition and a greater focus on students. Minimally regulated systems push managerial behavior in the direction of a private university model focusing more on tuition revenues, research income and research productivity. Their overall conclusion is that differences in political structures and economic conditions of states and regions play a more important role for performance than university autonomy. Controlling for such influences shows, however, that managerial autonomy influence choices of resource allocation and sources of revenue that influence performance.

Aghion and colleagues have conducted a number of studies on the relationship between organizational autonomy and performance (Aghion et al. 2007, 2008, 2009). Their outcomes suggest that university research performance is positively correlated with university autonomy, the level of funding and competition. Size (‘big is beautiful’) as well as age (‘reputation effect’) matters for research performance (i.e. position in the Top 500 of the Shanghai ranking). They also detect an interaction effect: higher levels of funding (i.e. higher budgets per student) have more impact when combined with financial autonomy. Their findings suggest a positive relationship between competition (for research grants) and research output (i.e. position on the Shanghai ranking). To test the causality of these relations, data for the U.S. were used and research performance was defined as the number of patents. From these U.S. data, the researchers “would (…) like to suggest” (Aghion et al. 2009: 24) a causal relationship between public university autonomy and competition on the one hand and research output in terms of patents on the other hand. As regards their findings for Europe, the study concluded, however, that there is not one single model for success in the university sector. The best countries in research performance in Europe, which include the UK, Scandinavian countries, the Netherlands and Switzerland, display quite some diversity in terms of both funding and university autonomy.

Van der Ploeg and Veugelers (2008) looked into the relationships between organizational autonomy, funding and performance of Europe’s universities. Primarily based on secondary analyses (using data from the Global Competitive Index of the World Economic Forum, the Shanghai and THES rankings, OECD indicators and the Bruegel group), they find some indications that a number of the better performing countries have high levels of autonomy for their universities while weak performing countries tend to be lower on autonomy. They found, however, prominent exceptions from this rule as well as a large dispersion of autonomy characteristics among high and low performers.

Recently, de Boer et al. (2010) and Jongbloed et al. (2010) investigated such links at the system’s level. They categorized the funding and autonomy situation of thirty-three European countries. For system performance eight dimensions were selected (access, mature learners, graduation, employability, international student mobility, research output, capacity to attract funding and cost effectiveness). For each performance indicator data for the years 1998, 2002 and 2006 were retrieved. In the analyses, the possible links between organizational autonomy and performance were controlled for the level of public investment in higher education (public expenditure on tertiary education as a percentage of GDP) and for the economic standing of the countries (on the Global Competitive Index, GCI). The studies draw the conclusion that within the right conditions (sufficient levels of public expenditure, financial incentives, sufficient capacity to attract and retain productive staff and sufficient capacity to meet demand) autonomous universities can contribute to improved research productivity in their countries as well as to higher levels of educational attainment. Organizational autonomy in academic staffing and related salary matters play an important role. Funding reforms (stronger emphasis on performance, more funding for research and the introduction of targeted research funding) and financial autonomy have contributed to an increase in research productivity in some countries. Funding reforms have also contributed to an increased number of graduates in some countries by providing incentives and autonomy to universities to grow. For other performance dimensions no significant link to university autonomy could be found.

Does autonomy matter? And for what?

Our review of the literature provides various important implications for future research. First, studies on the link between organizational autonomy and performance in higher education are surprisingly scarce. The evidence is inconclusive, while the studies point to organizational autonomy being one of the potential determinants of university performance. Further studies are thus needed to enhance our conceptual and empirical understanding.

Second, various promising propositions for future research can be condensed from the studies presented above. Organizational autonomy appears to be one predictor of the quantity of output of the primary processes of universities in research and teaching (e.g. numbers of graduates, articles published, number of patents, success in research grants). As regards the autonomy situation of the university, university’s decision-making capabilities to manage financial affairs as well as human resource affairs seem to be more important variables than other dimensions of autonomy. Interaction effects are to be expected, such as higher levels of funding having more impact when combined with autonomy in financial affairs, or the interaction between organizational autonomy and the type and level of competition within the field. Other variables need to be controlled for in testing the link between autonomy and performance. This includes system variables (wealth of nation and funding for higher education and research as an important—if not the most important—predictor of performance, competition within the field,) as well organizational characteristics (size, age, and organizational budget) that various studies have pointed at.

Evidently, the empirical evidence is inconclusive. The findings provide, however, some support for thinking guided by an institutionalist economic logic, namely that more autonomous universities that need to compete for resources are more productive. Especially in research, governmental control is likely to produce ‘government failure’ and will be less effective than making more autonomous universities with managerial discretion in prime production factors compete for resources and performance. Some support can also be found for an institutionalist sociological logic, namely that context matters, that organizational characteristics matter, and that there is not one single model for high performance or improving performance of universities to be expected.

Third, overcoming certain biases and limitations would greatly enhance our understanding of the predictors of university performance. The studies measure organizational autonomy in very different ways; most studies rely on data on the formal autonomy situation and do not distinguish between formal and real autonomy. Not (only) formal autonomy but the way autonomy is put into action might explain performance. One of the reasons for the inconclusiveness of the studies on autonomy and performance could thus be related to problems on the side of the ‘independent variable’. Currently, our knowledge is also biased towards certain indicators of universities’ research performance (except de Boer et al. 2010; Jongbloed et al. 2010) as the ‘dependent variable’. This situation is problematic for conceptual, normative and empirical reasons.

Just as autonomy, performance is a multidimensional concept. Performance is related to missions, goals or purposes that will affect the what and how to be measured, as well as its interpretation. Higher education systems are characterized by task complexity, multiple and possibly conflicting goals. And various stakeholders have various legitimate expectations and interests with respect to the performance of the university. In modern mass higher education, universities and other higher education providers also are expected to develop specific profiles. Thus, ‘one size does not fit all’ if specialization is taken to be seriously, and performance measures need to be aligned with organizational purposes and goals (van Vught 2009).

Further, the study of Knott and Payne (2004) implicitly points to possible trade-offs between different tasks and dimensions of performance (e.g. more autonomous universities shifting attention from teaching to research). Another trade-off concerns the relationship between quantitative output and quality of output. The data discussed are not inconsistent with the idea of publication inflation, that is universities having incentives to increase publication output irrespective of their impact in the field. The same holds for grade inflation, that is, universities might have incentives to keep students enrolled as long as possible if they are rewarded on enrolment and increase graduation rates if funding is based on diplomas. Our studies will only be able to account for such trade-offs by applying various dimensions of performance.

As with the concept of organizational university autonomy, the discussion above points to the need for context-sensitive and multidimensional concepts and measures of performance. Hence, we argue that a multi-dimensional performance concept as it is used in some ranking instruments, such as U-Multirank (van Vught and Ziegele 2012), is a good starting point. The dimensions used are education profile, student profile, research involvement, knowledge exchange, international orientation and regional engagement. This multi-dimensional perspective is stakeholder driven, accounts for different profiles and missions of universities, and supports the concept of multiple excellences. Such a multidimensional perspective would increase our knowledge on the contested relationships between autonomy and performance as two core concepts in contemporary higher education debates.

Conclusion

The main aim of this paper has been to contribute to the understanding of organizational autonomy and control in higher education reform and related expectations as regards the performance of universities. We discussed how the dominant narrative of political reform moves away from traditional beliefs in university autonomy that are built on institutional trust and linked to professional autonomy. The erosion of the traditional ‘social compact’ reflects deeper contextual and political changes as regards the role of higher education. In the emerging narrative of political change, autonomy becomes re-defined as the ‘new organizational autonomy’ of universities as both strategic actors and as an addressee of governmental control.

Our further empirical analyses was looking at the influence of principal-agent theory inspired reforms and built on structural-instrumental and cultural perspectives on formal and real autonomy applied to the Dutch case. The analyses highlights the influence of principal-agent thinking on regulatory policies as well as politico-administrative path-dependencies of reforms and compatibility problems within the field. The current political compromise reflects an ‘autonomization’ of the university that is built into regulatory policies for control that are expected to steer organizational behavior and to constrain organizational choices. Dutch universities have gained in managerial autonomy while they have lost in institutional autonomy. Governmental re-regulation has also installed a multi-level and multi-actor playing field of autonomy and control. Unintended effects and insufficient outcomes of policies of the past stimulate further policies for bonding universities to the government and getting governmental incentives right.

We thus conclude that notions of ‘state supervision’ are misleading insofar they suggest that government is stepping back from control in higher education. Notions of ‘autonomy and accountability’ are equally misleading insofar they suggest that performance monitoring is the price paid for more autonomy. Dutch universities are nowadays confronted with a new regulatory regime of control in which their managerial autonomy is supposed to be instrumental in aligning them to governmental goals and performance expectations. This is what we call ‘regulatory autonomy’.

But does organizational autonomy matter for performance? Our review of international studies on the link between organizational autonomy and performance reveals some promising though preliminary conclusions. Expectedly, financial resources (wealth of nations, funding for higher education, budgets of universities) turn out to be a strong predictor of the quantity of output of the primary processes of universities. University’s managerial decision-making capabilities (financial and human resource affairs) appear to be a co-predictor. Interaction effects have been observed with the degree of competition within the field as well as with certain organizational variables (size, age). Evidently, our review of the empirical literature provides some support for the institutionalist economic logic, namely that more autonomous universities with managerial discretion that need to compete for resources are more productive. Variations and exceptions are, however, hard to account for. More in line with sociological approaches in the study of public agency, no one single regulatory model for success of universities has so far been detected. The empirical evidence is inconclusive, partly because measures of university autonomy vary and mostly address the formal autonomy situation. Performance measures tend to be biased towards quantities in research output; and hardly account for the multiple-product character of universities, various organizational missions, and effects on the quality of services provided.

Finally, and since our evidence is limited, we want to point to certain avenues for further research on ‘regulatory autonomy and performance’. Our framework on organizational autonomy and control in higher education could be applied in cross-national and cross-organizational comparative perspectives. A comparative analyses of the multi-dimensional formal autonomy situation of universities would allow for further explanatory analyses of the local adoption of changing regulatory policies and their effects on the real autonomy situation of the universities. Further empirical investigation would thus address the analyses of university actor’s perceptions of their autonomy situation and for enhancing our understanding of ‘autonomy in use’. Opening the ‘black box’ of universities as organizations would also greatly enhance our insight into the interaction between the external and internal control of universities. Finally, context sensitive and multidimensional performance measures would allow to further test the political theory of practice that assumes that a certain regime of autonomy and control of universities would improve actual performance. After all, we want to know if governance matters, and studies on the determinants of universities’ performance are surprisingly scarce. In this respect, multidimensional approaches on performance are needed.